Andrew Ng 's machine learning lecture note (7)

Multi-class classification

When facing the classification problem, the output are multiple values , we call this multi-class classfication problem.In order to solve the problem, we use the algorithm 'one vs all'.

Suppose that we have 3 classes here. And we use circle, trangle, cross to represent those 3 classes . 

then we use the decision boudary to classify them into two parts (one part is for class i(i=1,2,3) the other part excluding class i ) ,then again we change the problem into logistic regression problem , here we get 3 hypothesis

 

In order to make the predition of x, we have this formular.